ifrcgo / dref-nlp Goto Github PK

Available apps:

• Parsing PDFs (dref_parsing)
• Tagging text (dref_tagging)
• Parsing + Tagging

conda create --name dref python=3.8
conda activate dref
python -m pip install -r requirements.txt
python -m spacy download en_core_web_md
python -m pip install -e ./dref_parsing
python -m pip install -e ./dref_tagging

To run the dref_tagging app (or the joint app) you must download the file DREF_docBERT.pt and add it to the dref_tagging/dref_tagging/config file. Currently to download this file you must contact the IM team at IFRC

The apps are made using fastapi and can be started by commands

uvicorn dref_parsing.main:app --reload
uvicorn dref_tagging.main:app --reload
uvicorn main:app --reload

and then opening in a browser the indicated web-page, usually http://127.0.0.1:8000/docs

The current version of the apps is available at: https://drefnlpdev.azurewebsites.net/docs

It contains the joint apps (for parsing the data and tagging it).

If the dref_tagging is needed (without parsing the data), it is available at: https://dreftagging.azurewebsites.net/docs

The docker images uploaded to Azure were created by running the following build command from the folders dref_tagging, dref_parsing or the root folder:

docker build -t myimage .

To run the docker locally, use

docker run -p 8000:8000 myimage

and open http://localhost:8000/docs